u10_rms vs u10_mean in gas flux parameterizations

[klindsay28]

The Wanninkhof 1992 piston velocity uses u10^2.
This is used for O2 and CO2 gas flux parameterizations.
So we have had the flux coupler pass the time mean of u10^2 to POP.

The newer NH emission code uses a linear combination of u10 and u10^2.
The initial implementation sets u10 to the sqrt of the time mean of u10^2, i.e. it is u10_rms.
This is different from the time mean of u10. It is not clear which is appropriate.

If we want to use the time mean of u10, then we need to pass that from the flux coupler to POP and add u10 to the MARBL interface.